Human tooth | Homo

We recently were able to publish a study where we automatically segmented the root canal space in human teeth: doi:10.1186/s12903-021-01551-x

Here’s a little explanation of the study, in which we used the sample changer of our SkyScan 1272 quite intensively. In collaboration with the zmk bern – Zahnmedizinische Kliniken we were asked to image +100 human teeth (excess material from tooth operations/extractions) and prepare the tomographic datasets for characterization. As mentioned, we used the sample changer of our scanner for this, on which we mounted each tooth in a custom-designed and 3D-printed sample holder. After aqcuiring +800 GB of raw data we reconstructed the data into about 350 GB of reconstructions. All these reconstructions were then ingested into a pipeline in a Jupyter notebook with Python code. The notebook is freely available here: https://github.com/habi/zmk-tooth-cohort and is able to be run with binder by clicking the relevant button in the GitHub repository above.

After we’ve prepared the data for our collaborators we’ve tried to write up the whole process as a publication. This was achieved by using the collaborative writing function of Manuscripts.io. A ready to submit-version of the manuscript was exported to a set of Markdown files and imported into a manubot-enabled repository on GitHub, which automatically converts the text into both a nice-looking website and PDF file.

The tooth shown in the header image of this post was scanned on a SkyScan 1272 with a source voltage of 80 kV, a source current of 125 µA with camera and geometry settings leading to an isotropic voxel size of 10 µm. The whole scan of this tooth took about 3 hours and 15 minutes. The animation was generated with MeVisLab using the MeVis Path Tracer feature. The movie was generated from a set of single frames in the same manner as described previously.